Should You Offer a Money-Back Guarantee on a Claude Code Build?
- A blanket 'not happy, full refund' guarantee is a bad fit for a build, because the hours are already spent and code cannot be un-shipped the way a physical return can be. It invites the wrong kind of prospect and creates an open-ended liability you cannot actually price.
- A narrow guarantee tied to a specific, defined checkpoint, usually the first milestone or the discovery deliverable, closes the same trust gap without the open-ended downside. It is a promise about the start of the engagement, not the whole thing.
- The prospects asking for a guarantee are almost never haggling on price. They are asking whether you are safe to say yes to. Answer that question directly and the guarantee itself barely needs to be generous.
The Direct Answer
Offer a guarantee, but never the blanket kind. A guarantee scoped to the first milestone, the discovery phase, or a specific defined deliverable does the actual job a guarantee is supposed to do, which is remove the fear of hiring a stranger, without exposing you to a refund request three weeks into a build when most of the hours are already spent. A full-project 'money back if you are not satisfied' guarantee sounds generous and is, in practice, a promise you cannot responsibly keep.
The Real Reason Prospects Ask for One
When a prospect asks about a guarantee, they are rarely negotiating price. They are asking a version of one question: what happens if I pay you and this does not work out? That is a trust question, not a discount request, and it shows up hardest with a client who has never hired an independent builder before, or one who has been burned by a freelancer who took a deposit and disappeared.
Answering it with a price concession misses the point entirely. What the prospect wants is evidence that you have thought about the failure case, not a lower number. That distinction is the same one that runs through [presenting your price without flinching](/blog/present-your-price-without-flinching): the confident move is naming the risk and how it is handled, not softening the number to make the risk feel smaller.
Why a Blanket Guarantee Is the Wrong Instrument
A retail return works because the store still has the product. A blanket money-back guarantee on a build does not work the same way, because by the time anyone is unhappy enough to ask for a refund, your hours are already spent and cannot be recovered. You are not offering to take a product back, you are offering to work for free and then also refund the client, which is a much worse trade than it sounds on a sales call.
It also selects for the wrong buyer. A client shopping for the safety net of a full refund on a custom build is often a client who has not fully committed to the project succeeding, and a small number of them are shopping specifically for builders naive enough to offer one. Neither is the client you actually want, and the pattern is close enough to the objection covered in [handling that's too expensive without dropping your price](/blog/handling-thats-too-expensive) that the same instinct applies: a concession aimed at the wrong worry attracts more of the wrong worry.
The Narrow Guarantee That Actually Works
Scope the guarantee to a single, defined checkpoint, not the whole engagement. The version worth offering: if the first milestone, or a paid discovery phase, does not land as agreed, the client is refunded that portion and the engagement ends there. Nothing beyond that milestone is covered, because nothing beyond that milestone has happened yet.
This works because it is a promise about the part of the relationship where trust is thinnest, the first payment before either side has proven anything, and it is bounded on both ends. You know exactly what it can cost you, because it can never exceed the value of that first milestone. The client gets exactly what they were actually worried about, which is not losing everything to someone who takes the money and vanishes.
- Define the milestone in writing before the guarantee means anything, the same document referenced in [the one-page agreement every builder should send before starting](/blog/one-page-agreement-before-you-build).
- Attach the guarantee to a deliverable, not a feeling. 'The discovery document is not delivered as scoped' is enforceable. 'You are not happy' is not.
- Cap it explicitly. State the guarantee applies to the first milestone only, so there is no ambiguity about whether it extends to work delivered in week four.
- Pair it with the deposit structure you already use, which is the natural place to state it, covered in [the deposit conversation: get paid upfront without awkwardness](/blog/the-deposit-conversation-get-paid-upfront).
How to Word It So It Closes Deals Instead of Attracting Refund-Seekers
Wording matters more here than in almost any other part of a proposal. A vague guarantee reads as marketing and gets skipped. A specific one reads as a real term and gets remembered at exactly the moment a nervous buyer is deciding whether to sign.
| Weak wording (avoid) | What it actually promises |
|---|---|
| "100% satisfaction guaranteed" | An undefined standard, decided entirely by the client, at any point in the project |
| "Money back if it doesn't work" | No defined checkpoint, so the liability grows the longer the project runs |
| "[Milestone name] delivered as scoped in the agreement, or that portion is refunded and we stop there" | A specific, boundaried promise tied to one deliverable and one payment |
Guarantee wording that helps you vs. wording that exposes you
Say it out loud on the sales call before the client asks, not as a defensive answer to a pushback. Naming your own guarantee unprompted is a stronger signal than confirming one when pressed, which is the same close-first instinct behind [the demo-first close](/blog/the-demo-first-close): showing confidence before it is demanded reads very differently from producing it under pressure.
When Not to Offer One at All
Skip it on very small or very cheap engagements, where the milestone and the whole project are close enough to the same thing that a milestone guarantee is functionally a blanket one again. A guarantee on a two-day fixed scope has nowhere useful to be scoped to.
Skip it, or scope it tighter, with a prospect who is already negotiating hard on other terms before the contract is even signed. A guarantee is meant to lower the temperature on a legitimate first-time-hire nervousness. It is not meant to be one more term in a negotiation with someone who is testing how much you will concede before the work has even started, a pattern related to the scope-creep instincts in [ten ways builders leave money on the table](/blog/ten-ways-builders-leave-money-on-the-table).
A Script for Presenting It on a Sales Call
Introduce it plainly, after the price and before you ask for the decision: 'The first phase is the discovery and plan, priced at [amount]. If that does not land the way we scoped it here, you get that back and we part ways there, no hard feelings. Everything after that is priced separately once we both know exactly what we are building.'
That sentence does three things at once. It answers the unspoken trust question before it is asked, it draws a clean line around what is and is not covered, and it reframes the first payment as low-risk instead of a leap of faith, which is usually the actual thing standing between a good conversation and a signed agreement.
Common Mistakes
- Offering a full-project refund guarantee to close a hesitant lead, then discovering the liability three weeks later when most of the hours are already spent.
- Writing the guarantee in vague, feeling-based language ('satisfied,' 'happy with the results') instead of tying it to a specific, deliverable checkpoint.
- Only mentioning the guarantee when a prospect pushes back on price, instead of stating it upfront as a term of how you work.
- Applying the same guarantee terms to a two-day project and a two-month one, when only the second has enough distance between milestones for a narrow guarantee to make sense.
- Treating the guarantee as a substitute for the written agreement, instead of a clause inside it.
Inside the Claude Code Profit Room, members compare the actual guarantee language they put in front of clients and which version got the deal signed without ever getting invoked. It is $9 a month at https://www.skool.com/claudecodeprofitroom/about, come in with a proposal you are about to send and we will look at the wording with you.
Short, practical drops on offers, outreach, pricing, and closing clients with Claude Code. No spam, unsubscribe anytime.
Frequently asked
Should a Claude Code builder offer a money-back guarantee?
Yes, but scoped narrowly, not as a blanket 'not satisfied, full refund' offer. A guarantee tied to a specific first milestone or discovery deliverable answers the trust question a prospect is actually asking, without exposing you to an open-ended refund liability once most of the project's hours are already spent.
Why is a full-project refund guarantee a bad idea for custom builds?
Because the labor is already spent by the time a client would invoke it, code cannot be returned the way a physical product can, and the liability grows the further into the engagement the request comes. It also tends to select for clients who are not fully committed to the project succeeding, rather than reassuring the genuinely nervous first-time buyer it is meant to help.
What should a narrow money-back guarantee cover?
One defined checkpoint, most often the first milestone or a paid discovery phase, written against a specific deliverable rather than a feeling. If that milestone is not delivered as scoped in the written agreement, that portion is refunded and the engagement ends there. Nothing after that checkpoint is covered, because nothing after it has happened yet.
When should I mention the guarantee on a sales call?
State it upfront, right after you give the price and before the client has to ask, rather than saving it as a defensive answer to a price objection. Naming your own guarantee unprompted reads as confidence. Producing one only when pressed reads as a concession, and concessions made under pressure tend to invite more pressure.
Last reviewed August 28, 2026.
Co-founder of the Claude Code Profit Room. Built and sold AI services to real clients; writes about offers, pricing, outreach, and closing with receipts.
More from Duncan Rogoff →
